Messages in this thread | | | Date | Wed, 20 Nov 2013 11:08:54 +0100 | From | Michal Hocko <> | Subject | Re: [to-be-updated] mm-memcg-add-memoryoom_control-notification-for-system-oom.patch removed from -mm tree |
| |
On Wed 20-11-13 00:05:18, David Rientjes wrote: > On Tue, 19 Nov 2013, akpm@linux-foundation.org wrote: > > > Subject: [to-be-updated] mm-memcg-add-memoryoom_control-notification-for-system-oom.patch removed from -mm tree > > To: rientjes@google.com,hannes@cmpxchg.org,kamezawa.hiroyu@jp.fujitsu.com,mhocko@suse.cz,mm-commits@vger.kernel.org > > From: akpm@linux-foundation.org > > Date: Tue, 19 Nov 2013 13:38:14 -0800 > > > > > > The patch titled > > Subject: mm, memcg: add memory.oom_control notification for system oom > > has been removed from the -mm tree. Its filename was > > mm-memcg-add-memoryoom_control-notification-for-system-oom.patch > > > > This patch was dropped because an updated version will be merged > > Why is this removed?
I've asked Andrew to drop it for now (you were CCed) and mentioned my reasons.
> I've laid out my perspective for doing userspace oom handling and this is > a vital part for system oom handling. I know that we are currently > discussing alternative proposals but my proposal is by far the most > complete and allows the most powerful policies to be implemented in > userspace. I'd prefer if we would keep the patch unless an actual > alternative is proposed and agreed upon, we can still discuss alternatives > while this in -mm and this patch in no way precludes other mechanisms from > being implemented. > > So why remove this?
This is a user interface visible change. I do not want to do it until we agree on a way to go. I do not see any advantage of having this in -mm until then. It doesn't need any testing from -next (does it?) and the code is simple enough to push it later on without troubles if the memcg.oom_control is considered as a way to go. I do not see any reason to rush it in now.
-- Michal Hocko SUSE Labs
| |